wewbull

joined 3 years ago
[–] wewbull@feddit.uk 1 points 4 hours ago

Well...a condom would let nothing through.

Are you saying you're a broken condom?

[–] wewbull@feddit.uk 2 points 2 days ago

At least this is money primarily going to UK workers.

Trident pisses me off because it buys missile we're perfectly capable of designing and making ourselves. Relying on another nation for last resort national defence is bonkers if you have the choice.

[–] wewbull@feddit.uk 1 points 2 days ago

Isn't that 2020 thinking?

Ukraine has show how they are vulnerable.

[–] wewbull@feddit.uk 9 points 2 days ago (2 children)

It's Trident.

Andy Burnham is expected to unveil on Thursday an £8.4bn investment for the UK's next generation of nuclear submarines.

Aimed at preparing Dreadnought-class boats for service by the 2030s, the nuclear-armed vessels are set to replace the current Vanguard-class and will be equipped to launch Trident 2 D5 ballistic missiles

This is on top of £16B since the 2020 defence spending report, and £30-40B before that (Wikipedia is a bit fuzzy on some of it). This is sunk-cost spending (no pun intended). There's no other option except cancel the program.

...and none of it covers the missiles, which isn't an independent system.

We need to have much better government procurement oversight. Contract values need to be understood properly upfront. Bids are intentionally low, with companies knowing that once they're "in" costs can go up and they'll get paid. It's a well understood strategy, but it continually happens.

[–] wewbull@feddit.uk 1 points 3 days ago

The problem with slop code is not superficial.

The game the machine is playing is " reproduce text that looks like all the other text you've seen before". As such the code will look good, read well, etc. The problem is that the problems are hidden. Incongruent logic between parts of the same program. Subtle "unintentional" behaviour. Bug fixes which force an outcome rather than remove a fundamental issue.

Apart from excessive verbosity it can be hard to see the issues.

[–] wewbull@feddit.uk 0 points 3 days ago

I always explain this "nudification" process as...

You know how if you put black bars over someone in a bikini or speedos, and suddenly they look naked? You haven't seen the person naked, but you've got a good idea what they look like naked because your brain filled in the gap. It's exactly what the diffusion model is doing.

I'm not sure it's realistic to stop the models doing it entirely given how they work. If you want it to fill in the background when you cut your ex out of a photo, it's going to fill in any other gaps.

[–] wewbull@feddit.uk 1 points 3 days ago (1 children)

In this case it's an advantage. If your enemy has a blind spot you exploit it.

 

Okay, we have a new contender for Most AI Thing to Ever Happen

  1. July 25th: someone messes around with an LLM and posts a proof of the Collatz conjecture that does, in fact, verify in the theorem prover. (The AI use is not disclosed on the github page) https:// github.com/xrchz/CollatzLean

  2. July 26th: several serious bugs are posted in the theorem provers, that in principle could allow a false statement to be "proven" true. They're serious, yes, but no need for panic, because you're not going to blunder into accidentally exploiting the bugs while writing a proof, probably. https:// github.com/leanprover/lean-kernel-arena/pull/81

  3. July 28th: someone who was right to be very skeptical of the Collatz proof, and had the expertise to study it with a fine-toothed comb, discovered it was exploiting a bug https:// github.com/leanprover/lean4/issues/14576

  4. The "proof" turns out to be exploiting multiple similar but distinct bugs to pass different solver variants!

  5. the human who posted the proof acknowledges the AI use and claims they did not knowingly point it towards the bugs it exploited. https:// leanprover.zulipchat.com/#narrow/channel/270676-lean4/topic/Counterexample.20to.20the.20Lean.20Conjecture.20.28Soundness.20Bug.29/near/613135216

Note that the proof was posted shortly before the related bug reports were posted. It is an open question if the AI found people discussing the bugs shortly before they were formally posted and "decided" to exploit them, if the AI "knew about it" as a learned strategy from the training stage (putting every single "proof" it's ever made and ever will make into profound doubt), or if it's recently been repeatedly blundering into it by sheer stupidity and that's how people noticed the bug at about the same time.

Theorem provers aren't magic, and have bugs just like all other programs. They are tools to help us double-check our reasoning. When you skip the reasoning and ask an AI to "prove" something for you that's over your head, you're entering an adversarial pact with the monkey-pawed Devil of Customer Satisfaction.

[–] wewbull@feddit.uk 14 points 4 days ago (4 children)

A lot of people don't realise SKG is a foothold in exactly that fight. If you establish precedent on something as "frivolous" as video games you can use it on much higher value thing too.

[–] wewbull@feddit.uk 29 points 4 days ago (5 children)

This is what the GPL was meant to fight against. Having boxes running software that is still under the control of the manufacturer.

Instead people complained that it was hard to use the code at work. YES. THAT'S THE POINT.

[–] wewbull@feddit.uk 3 points 6 days ago (5 children)

My point was that they fabricated a pretense to avoid needing one, just as you said. You're making the same fucking point.

[–] wewbull@feddit.uk 2 points 6 days ago (7 children)

Did you stop reading at that point?

 

Great example of why using placeholders like xxx@deleteduser.com is a bad idea.

view more: next ›